Unsupervised Learning Overview

The probability of event A happening given that event B has already happened.

The ratio of the observed probability of two events occurring together to the expected probability if they were independent.

The proportion of transactions containing both item X and item Y, divided by the proportion of transactions containing item Y.

The probability of item X being purchased given that item Y is already in the basket.

An algorithm that generates item sets (frequent itemsets) from transactional data.

A straight-line distance between two points in Euclidean space.

The sum of absolute differences between coordinates of two points.

Measures the angle between two vectors, indicating similarity or dissimilarity.

A method for determining the optimal number of clusters in K-means clustering.

The sum of squared distances between each data point in a cluster and its centroid.

A point that represents the center of a cluster in K-means clustering.

The process of assigning each data point to the closest cluster based on its distance to the cluster's centroid.

The process of repositioning the cluster centroids to the actual mean of the assigned data points.

This clustering technique uses probability distributions to create groups based on data similarities, like grouping words based on properties like "man's shoe" and "woman's glove" into categories like "shoe" and "glove" or "man" and "woman".

A machine learning algorithm used for grouping data points into clusters, where each cluster represents a set of data points with similar characteristics.

The process of identifying hidden patterns and structures within datasets.

Utilizing insights gained from clustering analysis to build predictive models that classify data into predefined categories.

The distance between two data points, often calculated using the Euclidean distance formula.

A metric used to assess the overall value of a customer to a business, considering factors like purchasing history, frequency, and average order value.

The process of determining the optimal number of clusters for a dataset, considering factors like minimizing within-cluster variance and maximizing between-cluster variance.

A type of machine learning where the computer learns from unlabeled data to discover patterns and insights without explicit instructions.

A technique for grouping data points into clusters based on their similarity. Items in a cluster are more alike to each other than to items in other clusters.

Rules that describe relationships between items in a dataset, often used to find patterns or associations in sales data.

This type of clustering allows a data point to belong to only one cluster. Think of it as putting items in separate, distinct boxes.

This type of clustering starts with each data point as its own cluster, then iteratively merges the closest clusters together.

This type of clustering allows a data point to belong to multiple clusters with varying degrees of membership. Think of it like a fuzzy category where items can overlap.

K-means clustering is considered converged when the cluster assignments no longer change, indicating the algorithm has reached a stable state.

One advantage of K-means is its ease of implementation. The algorithm involves straightforward steps that can be readily coded.

K-means can easily assign new data points to existing clusters by measuring their proximity to the cluster centers. This is useful for categorizing new data.

Categorical variables represent distinct categories (e.g., colors, genders) and are not easily handled by K-means, which is primarily designed for numerical data.

The initial placement of cluster centers can significantly influence the final clustering results in K-means. Choosing a good starting point is crucial.

The Apriori algorithm uses a prior knowledge principle, assuming that all subsets of a frequent itemset must also be frequent. It leverages this knowledge to efficiently discover frequent itemsets in datasets.

The Apriori algorithm follows an iterative approach, building on the knowledge of frequent itemsets of a given size (k) to find frequent itemsets of the next size (k+1).